@nikhilayeturi23/rltool

A ready-to-use React hook for AI-powered text optimization using reinforcement learning. No backend setup required!

What is this?

This package provides a React hook (useRLTool) that instantly adds Q-learning based optimization to your React app. Perfect for:

• 📧 Email tone conversion - Casual → Professional
• 📝 Content generation - Marketing copy, blog posts
• 🔧 Data normalization - Clean and standardize data
• 💬 Customer support - Response templates
• 📊 Text transformation - Any text optimization task

Zero backend setup required - uses our hosted RL optimization API powered by Cloudflare Workers and OpenAI

📚 Documentation

• README.md (this file) - Complete API reference
• CHANGELOG.md - Version history

For more examples, visit the GitHub repository.

Installation

npm install @nikhilayeturi23/rltool

Quick Start - 2 Minutes!

1. Install the package

npm install @nikhilayeturi23/rltool

2. Use in your React component

import { useRLTool } from '@nikhilayeturi23/rltool';

function EmailOptimizer() {
  const { optimize, loading, result } = useRLTool();

  const handleOptimize = async () => {
    await optimize({
      userQuery: "hey send me that file yo",
      objective: {
        goal: "Make professional and polite",
        constraints: {
          mustInclude: ["please", "thank you"],
          mustAvoid: ["yo", "hey", "slang"],
          tone: "professional"
        }
      },
      useCase: "email"
    });
  };

  return (
    <div>
      <button onClick={handleOptimize} disabled={loading}>
        {loading ? "Optimizing..." : "Make Professional"}
      </button>

      {result && (
        <div>
          <p><strong>Original:</strong> hey send me that file yo</p>
          <p><strong>Optimized:</strong> {result.optimizedOutput}</p>
          <small>Iterations: {result.iterations} | Reward: {result.finalReward}</small>
        </div>
      )}
    </div>
  );
}

Real-World Examples

API Reference

Hook: useRLTool(options)

Real-World Examples

Example 1: Email Tone Optimizer

import { useRLTool } from '@nikhilayeturi23/rltool';

function EmailOptimizer() {
  const [input, setInput] = useState("");
  
  // No apiEndpoint needed!
  const { optimize, loading, result } = useRLTool();

  const makeProf = () => optimize({
    userQuery: input,
    objective: {
      goal: "Make it more optimized",
      constraints: {
        mustInclude: ["greeting", "signature"],
        mustAvoid: ["slang", "abbreviations"],
        tone: "professional"
      }
    },
    useCase: "text"
  });

  return (
    <div>
      <textarea value={input} onChange={(e) => setInput(e.target.value)} />
      <button onClick={makeProf} disabled={loading}>
        Make Professional
      </button>
      {result && <p>{result.optimizedOutput}</p>}
    </div>
  );
}

Example 2: Data Normalization

const { optimize } = useRLTool(); // Uses hosted API

const normalizeData = async (rawData: string) => {
  await optimize({
    userQuery: rawData,
    objective: {
      intent: "Normalize user data for consistency",
      domain: "data",
      constraints: {
        mustInclude: ["consistent casing"],
        mustAvoid: ["duplicate entries", "mixed types"],
        tone: "technical"
      }
    },
    useCase: "data"
  });
};

Use Cases

This hook works for any text optimization task:

• ✅ Email Optimization: Casual → Professional, Angry → Polite

• ✅ Content Generation: Blog posts, product descriptions, social media

• ✅ Data Normalization: Clean inconsistent data, format standardization

• ✅ Code Documentation: Generate comments, API docs

• ✅ Marketing Copy: Headlines, CTAs, landing pages

• ✅ Customer Support: Response templates, help articles

• ✅ Translation & Localization: Adapt content for audiences

  TypeScript Types

All types are exported from the main package:

import { 
  RLOptimizationResult, 
  RLProgressLog, 
  RLObjective,
  OptimizationInput 
} from '@nikhilayeturi23/rltool';

Use Cases

This package works for any optimization task:

• ✅ Text Optimization: Email tone, marketing copy, code comments
• ✅ Data Normalization: Clean inconsistent data, format standardization
• ✅ Content Generation: Blog posts, product descriptions, social media
• ✅ Code Optimization: Refactoring suggestions, performance improvements
• ✅ Query Optimization: SQL, GraphQL, search queries
• ✅ Configuration Tuning: Finding optimal settings/parameters
• ✅ Creative Writing: Story generation, dialogue improvement
• ✅ Translation: Language translation with style constraints

Q-Learning Behind the Scenes

The hosted API uses these Q-learning parameters:

• Learning rate (α): 0.1 - How much new information overrides old
• Discount factor (γ): 0.95 - Importance of future rewards
• Epsilon: 0.2 (decays to 0.01) - Exploration vs exploitation
• Max iterations: 6 - Maximum optimization attempts
• Convergence threshold: 70 - Reward score to stop early
• Actions: refine_clarity, adjust_tone, add_details, simplify, restructure

Contributing

Found a bug or want to contribute? Open an issue or PR on GitHub.

Links

• 📦 NPM Package
• 🐙 GitHub Repository
• 🚀 Live Demo Worker